Breast cancer is a heterogeneous disease. A significant challenge in the development and testing of new therapies is determining which tumors a drug will target. Initial studies with a novel class of small molecule inhibitors of epigenetic pathways show therapeutic responsiveness in some breast cancers. Here, we propose to investigate the epigenetic changes that underlie diverse molecular phenotypes of breast cancer whose expression patterns have been previously considered on a global scale. We hope to identify which molecular phenotypes of breast cancer will be targeted by specific epigenetic pathway inhibitors. In this proposal, we will activate different epigenetic pathways in normal human mammary epithelial cells using adenoviruses carrying six different histone deacetylase (HDAC) or DNA methyltransferase (DNMT) family member genes to define six pathway expression signatures, characterized by system-wide patterns of activated and silenced genes. A set of over 1800 breast cancers, grouped by molecular phenotype (e.g. basal, luminal), will be queried for each of the six epigenetic pathway signatures to discover the specific epigenetic pathways that are deregulated in each phenotype. Next, we will determine which drugs best treat different phenotypes of breast cancer. Cell lines representing different molecular phenotypes of breast cancer will be treated with six different HDAC and DNMT inhibitors. The association of drug response to molecular phenotype in vitro will be used to predict tumor response in vivo. To do this, we will use fresh human tumors, classified by microarray analysis and orthotopically implanted in immunocompromised mice, to create a panel of tumor-specific xenograft models that represent the most aggressive molecular phenotypes of breast cancer: basal, ERBB2-overexpressing, and highly proliferating luminals. We will test the same six HDAC and DNMT inhibitors on the mouse models and measure how well they and a conventional chemotherapeutic agent (doxorubicin) inhibit tumor growth, verifying the in vitro predictions in an in vivo model system. We hope that our results will provide preclinical evidence for the initiation of a human trial that employs a more focused approach to the use of small molecule inhibitors of epigenetic processes by defining specific phenotypes of breast cancer that a given drug should target rather than testing that drug on all breast cancers. This proposal should demonstrate how tumor biology and genomic signatures can be translated into strategies for personalized cancer treatment.